Communication Dans Un Congrès Année : 2022

A computational model to account for conflict in moral and nonmoral decisions

Résumé

When facing a conflictual decision - like choosing between two desirable options - people tend to use simplifying strategies (e.g., preferring status quo) in order to deal with the complexity of the task and reduce cognitive load. This phenomenon should impact the decision-making process, from the way information is integrated to the weight given to considered evidence. However, a great majority of studies investigating this effect have been focused on nonmoral decisions, whereas we could expect different mechanisms in a moral realm. Moral decisions tend to be more extreme than nonmoral decisions, as morality often involves strong and clear-cut opinions. However, in the specific case of conflictual decision, extreme positions may be less psychologically comfortable to endorse and induce more nuanced responses. In order to investigate this effect, we built a computational model based on differential equations (dynamic neural fields coupled with sensorimotor control, extending classical drift diffusion models) in which a 1D population of neuronal units maps a continuous decision space. Neural fields usually operate on continuous spaces (e.g., sensorimotor), but allow the emergence of spatially localized attractors (which determine final decisions). In the decision space, activations of the corresponding accumulators account for the preference in the decision option continuum. Spatiotemporally coherent activity across the neural field reflects convergence in the decision space, and are able to model the non-linear dynamics of decision-making (e.g. measured in humans using mouse tracking techniques). We propose that conflict emerges when two regions of activation appear at qualitatively different places in the decision space (preventing the fusion of evidence into a consensual decision), and that activations are stronger for moral elements (leading to their prevalence in final decision). Simulated data with a priori defined parameters successfully reproduced empirical patterns, giving insights about what differs between conflictual moral and nonmoral decision processes and how different pieces of information (moral and nonmoral) are integrated during decision-making.

Fichier non déposé

Dates et versions

hal-04977914 , version 1 (05-03-2025)

Identifiants

  • HAL Id : hal-04977914 , version 1

Citer

Flora Gautheron, Jean-Charles Quinton, Annique Smeding. A computational model to account for conflict in moral and nonmoral decisions. International Choice Modelling Conference, May 2022, Reykjavik, Iceland. ⟨hal-04977914⟩
81 Consultations
0 Téléchargements

Partager

  • More